Mars transforms shop shelf-space insight with NTT DATA

Knowing your share of retail store shelf space in a fragmented market is critical to success in the dynamic and competitive world of fast-moving consumer goods. Capturing this use-of-shelf data is usually manual and labor intensive, with employees having to note what they see on shop shelves. With partner NTT Data and using computer vision, Mars in Italy is transforming how the data is recorded and used.

NTT DATA, leveraging its Milan Innovation Center with its replica supermarket, took the initiative to go to Mars and suggest a better delivery of a “planogram,” the retail industry term for visual representations of a store’s display of products—an essential tool in visual merchandising to maximize profits.

Shelf displays offer a new wealth of dynamic insight

Applying computer vision and artificial intelligence (AI), users can point a camera (in devices such as iPads) at the goods on shelves, capture images of their products, and spot and identify rival products, including price information and changes in share, price, and variety, from one week to the next. All this works offline, with AI running on edge—meaning there is no need for live connectivity. This new wealth of information means Mars can understand trajectories at a granular level and compare performance with rivals store to store and region to region.

One breakthrough implying computer vision is now a realistic solution for planograms is the reduced cost and effort required to train the AI. In the past, computer vision had to be applied to ingest images recorded from multiple angles to ensure the AI recognized a product in real-world scenarios. Adding a new product was a challenge every time, and new products came thick and fast in retail. NTT DATA’s application of generative AI (GenAI) in its product roadmap will substantially cut the training effort, reducing the requirement for sample images to a mere handful.

NTT DATA will make planogram innovation available to all

It takes brave clients and confident service providers to collaborate on shared innovations. The initial pilot proved the potential to both parties. The net result is that Mars now has access to industry-leading insight and is starting to roll out the benefits to other international markets. NTT DATA now has an asset for other clients to use.

The project, presented at NTT DATA’s European Analyst event, is just one of many cases the firm can showcase in which GenAI is an integral part of its product roadmap. It also illustrates the firm’s view that GenAI should not have its own strategy—it should simply be part of your business strategy. GenAI should be part of the solutions.

The Bottom Line: GenAI can make long-held dreams feasible when service providers step up and enterprises commit.

NTT DATA had worked with computer vision for a decade or more before developing this feasible planogram solution for fast-moving consumer goods. Applying GenAI’s ability to generate new data from samples makes the slow, expensive training of data sets fast and affordable. Enterprise leaders should revisit their backlogs and consider if some solutions they had considered too expensive or time consuming could become feasible through a similar application of GenAI.